Fix ranked point system

Games Halo Infinite
62

True dat.

So sometimes the best options is to simplify down to the core essence of what we are talking about.

So it makes perfect sense to reduce all the background noise. A lot of which is subjective. And focus on what really matters - were you good enough to win.

Your rank puts in you in the bottom 15% of the population.

And part of it sounds like it is your fault.

But you can’t really complain about the system itself if you are actually going out of your way to sabotage it.

If you want to rank up. Play smart. Stop dying.

MMR will only pull down your win if it’s lower than your CSR.

And if you keep self destructing with deaths - it probably is.

Wish I did.

But the truth is I just find ranking systems interesting. Kind of a psuedo stats / maths nerd.

Back in the good old days of CE and H2 we used to have regular LAN parties and we’d draw up our own tables. Trying to come up with better ways of ranking.

And I don’t think the system is flawless.

Far from it.

But TrueSkill2 itself is very good. And probably the best of any such systems.

I don’t agree with a lot of choices 343 have made. Particularly around the CSR system. And the fact that we don’t have an API to access our own information is a travesty.

I’m all for arguing how to improve the system. But mostly I just end up having to “defend” it from blatant mistruths that keep perpetuating. Largely because of 343’s hopeless communication.

63

My kpm is def not low if that’s the metric. Is that why I gain zero for ever beating anyone who’s diamond? Makes no sense.

64

It is what it is … my focus is winning. I only focus kills when in need point fixing. With the current system I have to win 75% of my matches to break even. Unless I do some kill stat work periodically because 5+5+5-15= zero. There’s no 2 ways around that. My last session brought my up to +7s and it was one day… Couple hours. For me it’s easy I caught it early but. You can technically win with zero kills since 80 percent of the games are objective based. The chances of that are like hitting the lottery jack pot twice. But it’s conceivable.

65

You’re obviously a pretty good player and no doubt you KPM would be good against an average player but is it higher than most people at your rank, while playing against similarly ranked opponents? If it is, then it should be driving your rank up, if not then you have probably reached a natural plateau where the game thinks it has you holding steady at the correct rank, based your current skill.

Unless you regularly outperform your peers in the rankings or perform above expectations in games against higher ranked opponents, you aren’t ever going to rank up significantly. You can grind out a few extra wins with a good team effort but unless those wins are in games you were expected to lose against, they just aren’t going to make a big difference.

1 Like
66

It’s always useful to look back on these things.

In your last five games you got 32 kills. The games lasted a total of 52 minutes exactly.

Which gives you a KPM of 0.6

And it’s amazing how the metric can get away from you if you are not specifically monitoring it.

Anecdotally you need a KPM of around 1.2 to hold your level (depending on the game mode). And there or higher to rank up. Josh used to talk about 1.3 to 1.5 as being an indicator of ranking up.

Maybe.

It depends on the make up of your squads and what the match maker can dish you up as opponents.

Try not to think of it as a certain W/L rate.

You just need to beat those sides ranked above you. And not lose to those sides ranked below you.

Do that and you will rank up regardless of the ratio.

And keep in mind that there is a squad dynamic here. In a way you are also competing with your team-mates for a slice of the ranking pie. And if you keep serving them up kills and dying more than needed - you are handing them a bigger slice of said pie.

The system is what it is. It may not be perfect for your modus operandi. But it seems like you wouldn’t have to make huge changes to milk it better.

Technically.

But you wouldn’t win many games doing that.

Play the objective AND get your share of the kills. No more. No less. Just your share.

KPM is only a weighting… but you are letting it drag you down.

67

No they didn’t. Don’t know where you got your info from but you couldn’t be more off. Running like 1.25-1.5

https://halotracker.com/halo-infinite/profile/xbl/CanadianBushMan/overview?experience=ranked&playlist=edfef3ac-9cbe-4fa2-b949-8f29deafd483

Tells you exact game times too if you look at them. No need to guess they always end early

68

Yep. Apologies. Clicked on the wrong account for HaloTracker.

That’s what I get for replying at work :slight_smile:

I don’t have exact numbers as we don’t know the weighting… and you can’t always tell exactly where your MMR stands relative to the opposition… but if you keep your KPM at least 1.2 then you should minimise any drops in MMR. If you can get it at, or above 1.5 then you will help to push your MMR up.

Unfortunately we’re walking in the dark here.

But, focus on winning. And while you are at it - keep an eye on your KPM. Don’t go looking for trouble - but if you can buff it a bit then you help nudge your MMR in the right direction. The danger comes when you start taking unnecessary risks.

I do it all the time. Chasing an opponent to try and nail the kill. But invariably you not only miss the kill - but you also get killed. And it’s very hard to buff your KPM while waiting to respawn. :frowning:

It’s a weighting - so focus on what’s important; Keep winning. Get your MMR up. Play better opponents. Win those. Rank up.

69

Still came out under a game but was 1.46 kpm. Above k/d etc. Didn’t make much of a difference. Did see a plus 8 one game but it’s been a while since I’ve gotten more the plus 7 iunno. Give it the long run I guess and see if it’s a sham wow

70

This is also very important.

In the days of 1-50 people would happily play for days, if not weeks, for their rank to change.

But now the scale is 1-1800+ we’ve resorted to losing our minds over 1-2 points at the end of each game.

Nobody cared if you went from 46.14 to 46.20 and your team mate went up by 0.02 more than you did. Because you didn’t know. We were all blissfully unaware of how the MMR oscillated in the background.

I swear it would be a lot better for everyone if the scale was reduced.

Is there really any difference between a CSR of 1234 and 1235? But people are losing their -yoink- over a point or two.

It shoudl be 1 to 100 (or even 117).

Let everyone sit back, relax, and let it unfold naturally.

71

I still have massive issues with it don’t get me wrong. Whoever thinks assists mean nothing especially in objective which is mainly what ranked is, is truly a -Yoink!- They are HUGE

72

They are not saying they aren’t.

If you do the right plays and get the right type of assists (ie. strategic team shooting) then that’s a huge leg up to the win.

And the win is what helps you to rank up.

There just isn’t any statistical evidence that ranking you up above and beyond the win is worth it. Either in increasing the accuracy of the system, or the increased data needed to use it.

If, as a team, you have good K/D, good objective time, a healthy number of assists, damage for vs against, etc - you have probably won the game.

If, as a team, you perform badly in those metrics - odds are you’ve lost.

So there is no point in double dipping and increasing (or decreasing) your rank twice.

Of course there will be the odd game where you do well in those metrics (or at least some of them) and still lose. But that’s just a blip in your journey.

And there will be soft games where you win despite doing little in those metrics.

But odds are, if you are playing a team ranked above you (which are the ones you need to win to rank up), then you will need to do all the right metrics well.

And the counterpoint is that if you are going to rank up on the basis of a metric, such as assists - you also need to rank down if you do poorly in that metric. Are you going to accept a brake on your rank if you don’t meet the expected number of assists?

So you have to reduce the rank change on the result and then add in weightings for all the relevant metrics - and there will still be people complaining that their particular metric of choice hasn’t been included.

So the simple thing to do is weight it all on the result and trust that over time it will all balance out.

73

I would literally accept a hit for not having enough assists. Doesn’t have to be a huge number but like a 6 threshold. Could be more but slayer prevents that. Even if the worth was half of what kills do, that’s more then enough. I’d love to see the diff in high ranked gameplay and team cohesion from the system now compared to a system that included that

74

You don’t have to.

Microsoft have already projected the rank changes if assists were used. They ran millions of Halo 5 games and used assists to recalculate people’s ranks.

And guess what. It wasn’t any faster or more accurate.

Don’t get me wrong. I’m a big fan of assists. When I play with my sons (better players) I take a lot of pride in assists. It’s my contribution. And as long as I minimise my deaths I can be part of the win.

But in terms of ranking - you would need a lot of processing resources to isolate good assists (team work) from bad assists (you lost that 1v1). And all for a metric that is marginal at best.

So just go with the win / loss.

Personally I would love to see 343 explore damage. Not just total damage - but subsections of damage;

  • total damage
  • damage for
  • damage against
  • initial damage (getting first hit in)
  • clean up damage (how much damage did you do for the kill).
  • damage that lead to a kill (the damage ‘assist’)
  • damage that was done but enemies escaped
  • damage that you took but escaped
  • damage over time
  • damage over distance
  • damage with each weapon (or grenades, melee, etc)
  • damage from different directions (eg. from the side, or behind)
  • damage footprint for (heat map) - both where you are and then where they are.
  • damage footprint taken
  • damage heights for and against

And so on.

You could actually paint a pretty accurate picture of someone’s game style from all the different break downs.

75

That would be pretty elaborate but yes I think that tells a good tale of the story for sure. I have a hard time believing the assists, but halo 5 was extremely different with just a one shot pistol for ppl. Less team fire went down in that game from my experience.

I’ll put it this way, in this game at this level I out before pretty well everyone 1v1 right now. If someone’s alone I’m 100% confident on challenging him because most times I’m getting him. But in most situations it’s team shooting because it’s never just one. So that’s why assists become so crucial. I’ll out br one person in hill and put a perfect 3 in another before dying, sacrificing my life for the team to break an obj. Setup. Things like that matter in a game bigggg time. Sometimes plays like that decide a game. Assists would matter. Someone going 15 9 and 3 isn’t better then a say 12 10 12. The 12 10 12 had a great game. 3 assists only did good because 12 10 12 bullied his way around helping the team (damage would probably show this though)

So I agree about the damage